ARTICLE

Received 19 May 2016 | Accepted 8 Nov 2016 | Published 7 Feb 2017 DOI: 10.1038/ncomms13876 OPEN Comparative influenza interactomes identify the role of 2 in virus restriction

Lingyan Wang1,*, Bishi Fu2,*, Wenjun Li3, Girish Patil1, Lin Liu1, Martin E. Dorf2 & Shitao Li1

Cellular protein interaction networks are integral to host defence and immune signalling pathways, which are often hijacked by viruses via protein interactions. However, the comparative virus–host protein interaction networks and how these networks control host immunity and viral infection remain to be elucidated. Here, we mapped protein interactomes between human host and several influenza A viruses (IAV). Comparative analyses of the interactomes identified common and unique interaction patterns regulating innate immunity and viral infection. Functional screening of the ‘core‘ interactome consisting of common interactions identified five novel host factors regulating viral infection. Plakophilin 2 (PKP2), an influenza PB1-interacting protein, restricts IAV replication and competes with PB2 for PB1 binding. The binding competition leads to perturbation of the IAV polymerase complex, thereby limiting polymerase activity and subsequent viral replication. Taken together, comparative analyses of the influenza–host protein interactomes identified PKP2 as a natural inhibitor of IAV polymerase complex.

1 Department of Physiological Sciences, Oklahoma State University, 264 McELroy Hall, Stillwater, Oklahoma 74078, USA. 2 Department of Microbiology & Immunobiology, Harvard Medical School, 77 Avenue Louis Pasteur, NRB830, Boston, Massachusetts 02115, USA. 3 School of Stomatology, Peking University, 22 Zhongguancun South Avenue, Beijing 100081, China. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to S.L. (email: [email protected]).

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876

nfluenza A virus (IAV) is a highly transmissible respiratory protein complex under IAV and mock infection conditions, pathogen and presents a continued threat to global health, respectively. To efficiently reduce false positives in AP-MS, we Iwith considerable economic and social impact1,2. IAV is a adopted the statistical method Significance Analysis of INTer- member of the orthomyxoviridae family and possesses eight actome (SAINT)11 in combination with the proteomic database segments of a negative-sense single-stranded RNA genome. derived from HEK293 cells in our laboratory (Supplementary During IAV infection, host pattern recognition receptors, such Data 1–3)12–14. SAINT allows bench scientists to select bona fide as TLR7 and RIG-I, sense viral RNA and elicit interferon- interactions and remove non-specific interactions in an unbiased mediated innate immunity to restrict IAV infection3. In addition, manner. Using a stringent statistical SAINT score (Z 0.89), we host intrinsic restriction factors impair IAV infection by identified 357 high-confidence candidate interacting interacting with viral proteins. For example, the E3 ligase (HCIPs) in HEK293 cells, which forms a PR8 IAV–host TRIM32 ubiquitinates PB1, thereby leading to PB1 protein interaction network of 529 protein interactions (Supplementary degradation and limiting viral infection4. By contrast, IAV Fig. 2 and Supplementary Data 1; note that the data set proteins engage with the host cellular protein interaction for C-terminal FLAG-tagged PB1 in HEK293 without viral network to hijack host molecular machinery to fulfil viral life infection was reported previously4). Our network features 243 cycle and perturb host defences to evade immune surveillance. IAV-induced interactions, 88 IAV-suppressed interactions and Thus, the protein interactions between IAV and host contribute 317 new interactions (Supplementary Data 4,5). to the outcomes of viral pathogenesis. Systematic comparison of multiple influenza–host protein inter- IAV comprises a plethora of strains with different pathogenic actomesisessentialfortheidentification of general mechanisms of profiles. Several recent proteomic studies identified a cohort of regulation of viral infection and the discovery of common therapeutic cellular factors that interact with IAV proteins5–8. However, targets. However, common and strain-specific virus–host protein the knowledge of common and strain-specific interactions is interactions have not been systematically investigated. We therefore incomplete and how these interactions control host defence and furthermappedtheproteininteractomesoffiveIAVproteins(PB1, viral infection remains to be fully elucidated. Systematic analysis PB2, NP, NS1 and M2) from two additional H1N1 strains (WSN/33, of strain-specific IAV–host protein interactomes should reveal A/WSN/1933 (H1N1); NY/2009, A/New York/1682/2009 (H1N1) general and distinct mechanisms of regulating viral infection and and one H5N1 strain (VN/2004, A/ /Viet Nam/1203/2004(H5N1)), host defence. Insights gained from these interactions will facilitate plus the protein interactomes of NS1 and NP from a H3N2 strain the design of future antiviral therapies. (Aichi, A/Aichi/2/1968 (H3N2)) in HEK293 cells (Supplementary Plakophilin 2 (PKP2) is the most prevalent plakophilin protein Data 6,7). NY/2009 is a pandemic strain, whereas VN/2004 is an and essential for the formation of desmosomes and stabilisation emerging highly pathogenic strain1,15. VN/2004 viral protein of cell junctions9. Mutations in the human PKP2 have been complexes were only purified from uninfected stable cell lines. linked to severe heart abnormalities leading to arrhythmogenic These viral proteins exhibited comparable expression levels in the right ventricular cardiomyopathy, an inherited disorder of the stable cell lines (Supplementary Fig. 3a–e). The cell viability and virus cardiac muscle10. However, the role of PKP2 in viral infection is growth of these stable cell lines were also comparable unknown. (Supplementary Fig. 4a,4b). These interactomes consist of 625 In this study, we first used a proteomic approach to establish a HCIPs, which is enriched in several pathways such as messenger comprehensive and dynamic interactome of 11 viral proteins of RNA (mRNA) splicing, RNA transport and virus modulation influenza A/Puerto Rico/8/1934 (H1N1) (PR8) in HEK293 cells. (Supplementary Fig. 4c). Approximately 30% (304 of 1014) of the Analysis of the network revealed that M2, PB1, PB2 and NP are interactions were previously reported (Supplementary Data 8), the major nodes connecting cellular factors with known and including well-established interactors such as PIK3R2 (ref. 16), predicted roles in immunity and viral infection. Thus, we further PPP6C17,POLR2B18 and IPO5 (ref. 19). Analysis of the interactomes mapped the protein interactomes of these 4 viral proteins plus revealed that the ratio of common HCIPs with each viral protein NS1, the multifunctional viral protein, from two other H1N1 (of Z2strains)variedfrom26to60%(Fig.1b).NP,PB1andPB2 strains and one H5N1 strain. In addition, the protein inter- shared more than half of HCIPs among strains, whereas NS1 and M2 actomes of NS1 and NP from a H3N2 strain were mapped. exhibited higher strain-specific interactions (Fig. 1b). Parallel comparisons of these interactomes revealed common and unique protein interaction patterns, suggesting general and distinct strategies of each viral strain. Gain- and loss-of-function Common and unique interaction patterns in the interactomes. studies of the common IAV interactors identified five novel host To comparatively analyse the interactomes, the relative protein factors regulating viral infection. Our study further demonstrates abundance of HCIPs in each complex was determined by that PKP2, a common PB1 interactor, inhibits the IAV normalized spectral abundance factor20. We first analysed NS1 polymerase activity, thereby limiting viral infection. protein complexes from PR8, NY/2009, WSN/33, Aichi and VN/2004. Surprisingly, NS1 proteins displayed distinct protein Results interaction patterns (Fig. 1c). For example, the well-known NS1 Mapping IAV–host protein interactomes. To uncover the interactors PIK3R2 and cleavage and specific comprehensive IAV–host protein interactions, we first performed factor 4 (CPSF4, also known as CPSF30) exhibited different affinity purification coupled with mass spectrometry (AP-MS) compositions in NS1 complexes. PIK3R2 accounted for 445% of analysis of PR8 IAV protein complexes (Fig. 1a). Eleven C- and NS1 complexes of PR8, NY/2009 and VN/2004 but only a small eight N-terminal FLAG-tagged viral were individually fraction of NS1 complexes of WSN/33 (B15%) and Aichi (B5%) transfected into HEK293 cells to make stable cell lines (Fig. 1c). By contrast, CPSF proteins and their complex (Supplementary Fig. 1a). The cell viability and virus growth of components (FIP1L1 and WDR33) were abundant in NS1 these stable cell lines were comparable (Supplementary complexes of WSN/33 and Aichi but absent in the PR8 NS1 Fig. 1b,c). Each stable HEK293 cell line was mock infected or complex (Fig. 1c and Supplementary Fig. 4d). These data are infected with 1 multiplicity of infection (MOI) of PR8 IAV consistent with the absence of F103 and M106 in NS1 of PR8 for 16 h. After infection, IAV protein complexes were affinity influenza, which is required for stabilising the CPSF interaction21. purified using anti-FLAG antibody and then analysed by mass The other pronounced interactors were PRPF39, prolyl spectrometry. Two biological repeats were obtained for each IAV 4-hydroxylases (P4HA1 and P4HB), and the tRNA-splicing

2 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ARTICLE

a

HEK293

NP NS1 NA M1 Viral gene FLAG

M2

SAINT PB1F2 LC-MS/MS analysis PA FLAG IAV HA Viral gene

PB2 PB1

NS2

b PR8 WSN/33 NY/2009 VN/2004 Aichi NS1 NP PB1 PB2 M2

6 11 24 1 1 1 0 17 55 25 62 1 0 1 0 57 57 6 0 1 2 1 18 19 3 2 9 1 5 4 4 1 9 6 1 0 2 26 34 1 2 0 3 7 2 2 18 4 27 28 2 5 35 1 4 2 29 1 2 1 29 37 22 8 8 36 9 1 0 3 0 29 3 4 2 2 6 2 2 1 5 1 1 2 0 3 0 4 3 2 3 2 0 100 10 38 6 19 0 3

Common 26% 60% 51% 58% 39% HCIPs c Common interaction Strain-specific interaction 0.45 PR8 PIK3R2 Uninfected 0.3 IAV RTCB HNRNPCL1 0.15 COPR5 DDX1 FKBP5 PFN2 0 0.45 NY/2009 PIK3R2

0.1 PIK3R3 GRB2 RTCB ACTN2 CPSF SKIV2L2 NKRF ZFC3H1 0 0.2 WSN/33 0.15 PIK3R2 CPSF FIP1L1 WDR33 0.1 P4HA1 P4HB PFDN2 RTCB DDX1 C14ORF166 0.05 TNFAIP8 0 0.2 Aichi 0.15 PIK3R2 PRPF39 FIP1L1 0.1 CPSF WDR33 TRNAU1AP P4HA1 P4HB 0.05 RTCB 0 0.5 VN/2004 PIK3R2

COPR5 0.1 FIP1L1 HIN1L PFKFB1 CPSF 0 d Common interaction Strain-specific interaction 0.15 PR8 MOV10 LARP Uninfected NARS2 0.1 HNRNPCL1 IAV DAP3 NOP10 SRP14 0.05 DKC1 KPNA SNRNP

0 0.15 NY/2009 QARS MOV10 0.1 HNRNPCL1 NARS2 PTCD3 RSL1D1 SNRNP70 OSBPL5 DAP3 DKC1 LARP NOP58 0.05 KPNA 0 0.15 MEPCE MOV10 WSN/33 0.1 LARP NARS2 NOP10 DAP3 HNRNPCL1 DKC1 SNRNP SRP14 0.05 KPNA 0 0.15 LARP Aichi MEPCE 0.1 HNRNPCL1 SNRPA NSAF of NP protein complexesDAP3 MOV10 NSAF of NS1 protein complexes SNRNP70 NOP58 SRP14 0.05 DKC1 KPNA NCBP1 ZNF846 0 0.15 LARP VN/2004 MEPCE 0.1 MOV10 DAP3 HNRNPCL1 SNRNP70 0.05 DKC1 NCBP1 NOP58 PRPF39 KPNA 0

Figure 1 | Comparative analysis of IAV–host interactomes. (a) Pipeline of AP-MS for mapping IAV–host protein interactomes. (b) The Venn diagrams illustrate the number of common HCIPs (in Z2 strains) in viral protein complexes from different IAV strains. The percentage of common HCIPs is indicated. (c,d) Comparison of the relative abundances of HCIPs of each NS1 (c) and NP (d) from five IAV strains. The outstanding HCIPs are labelled. NSAF stands for normalized spectral abundance factor.

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ligase complex consisting of RTCB (also known as C22ORF28), PKP2 restricts IAV infection. These initial observations led to DDX1 and C14ORF166 (ref. 22). In addition, each NS1 exhibited our working hypothesis that PKP2 may affect influenza viral unique interactors in the strain-specific interaction, such as replication. To test this hypothesis, four assay systems (reporter FKBP5, NKRF, TNFAIP8, ZFC3H1 and TRNAU1AP. assay, western blot, immunofluorescence and plaque assay) were In contrast to NS1, the interactomes of NP, PB1, PB2 and M2 used to evaluate the biological effects of PKP2 overexpression were more conserved among various IAV strains (Fig. 1d and on IAV replication. First, three FLAG-tagged armadillo repeat- Supplementary Fig. 5a–c). The NP interactomes featured containing proteins, PKP2, PKP4 (a.k.a. p0071) and b- common interactors involved in nuclear import (KPNA family (also known as CTNNB1) were transfected into HEK293 cells, proteins), transcription (LARP family proteins) and mRNA followed by infection with the PR8-Gluc. PKP2 but not the other splicing (SNRNP proteins). However, OSBPL5, a protein of armadillo repeat-containing proteins restricted IAV replication unknown function, was identified as a unique interactor in the (Fig. 4a). PKP2 exhibited minimal effects on the activity of a NP complex of NY/2009 (Fig. 1d). The PB1 and PB2 Wnt signalling reporter (Supplementary Fig. 8). Next, we interactomes shared several common interactors, such as examined whether PKP2 restricts other RNA viruses. HEK293 PRKAR1A and UBR5 (Supplementary Fig. 5a and b). cells expressing PKP2-FLAG were infected with IAV, vesicular Interestingly, PB1 was associated with DNAJC7, whereas PB2 stomatitis virus or Sendai virus. PKP2 selectively restricted IAV interacted with DNAJB3 or/and DNAJB6. The M2 interactomes replication (Fig. 4b). We then determined the effect of PKP2 included AGTRAP and EIF2B family members (Supplementary on viral infection by examining the NP protein expression of Fig. 5c). Thus, IAVs have common and distinct interactions with WSN/33 IAV. PKP2 overexpression resulted in viral restriction, the cellular protein network that may underlie the different as indicated by decreased levels of NP protein detected by western outcomes of viral pathogenesis. blot (Fig. 4c). Immunofluorescence assays were also performed to visualize viral restriction. Ectopic expression of PKP2 restricted IAV infection of HEK293 cells for multiple H1N1 strains (PR8, Identification of host factors regulating viral infection. WSN/33, NY/2009) and one H3N2 strain (Aichi) (Fig. 4d,e). To identify common host factors that modulate viral infection Plaque assays were used to determine the effect of PKP2 on the of different IAV strains, we established a ‘core’ interactome production of infectious IAV particles. Overexpression of PKP2 comprising 185 host factors associated with Z3 strains (Fig. 2a). consistently reduced viral titres in A549 lung epithelial cells We next examined the effects of 34 common interactors on viral (Fig. 4f). Taken together, these findings indicate that PKP2 is an infection by overexpression in HEK293 cells. After 24 h of IAV restriction factor. transfection, cells were infected with 0.1 MOI of a reporter virus, To complement the above overexpression data, we depleted the PR8 IAV carrying a Gaussia luciferase gene (PR8-Gluc)23 for PKP2 with siRNA. Decreased PKP2 protein expression was 16 h. We determined viral infection activity by luciferase assays correlated with increased IAV reporter activity in A549 lung cells (Fig. 2b and Supplementary Data 9). IAV infection was facilitated (Fig. 5a and Supplementary Fig. 9a). RNAi depletion of other by EIF2B4 (Fig. 2b). More importantly, four novel proteins PKP proteins or CTNNB1 had marginal effects on IAV infection (FKBP8, TRIM41, PKP2 and ZMPSTE24) inhibited influenza (Supplementary Fig. 9b), indicating the specificity of PKP2 on viral infection without pronounced impact on cell viability IAV restriction. To exclude off-target effects of RNAi, cells were (Fig. 2b and Supplementary Fig. 6a). The interactions between transfected with a siRNA resistant PKP2 rescue construct before these host factors and viral proteins were confirmed by infection with PR8 reporter virus. When combined with PKP2 co-immunoprecipitation (Supplementary Fig. 6b–f). RNA siRNA the rescue construct restored antiviral activity, validating interference (RNAi) depletion confirmed the requirement of siRNA specificity (Fig. 5b). In addition, silencing PKP2 enhanced EIF2B4 for viral replication in A549 and primary tracheal cells virus propagation of four IAV strains in A549 cells, as detected by (Fig. 2c). Furthermore, RNAi also confirmed the antiviral roles of immunofluorescence assays (Fig. 5c). Knockdown of PKP2 also PKP2, FKBP8, TRIM41 and ZMPSTE24 in A549 cells and enhanced IAV propagation in primary tracheal epithelial cells, as primary tracheal cells (Fig. 2c). Consistently, PKP2, FKBP8, detected by plaque assay (Fig. 5d). Therefore, endogenous PKP2 TRIM41 and ZMPSTE24 inhibited viral infection of NY/2009 and is essential for host restriction to IAV. WSN/33 (Supplementary Fig. 7a). RNAi knockdown efficiency and cell viability were verified (Supplementary Fig. 7b–c). Taken together, we identified five HCIPs that regulate viral infection. PKP2 inhibits IAV polymerase activity. Heterotrimeric IAV polymerase requires interactions between PB2 and the PKP2 interacts with PB1 polymerase. PKP2 is a scaffold protein C-terminus of PB1. PKP2 also interacts with the PB1 C-terminal for desmosomal cell–cell junctions, but the role of PKP2 in domain (Fig. 3c). These findings led us to examine whether PKP2 controlling IAV infection is unknown. Thus, we examined can compete with PB2 for PB1 binding, thereby limiting IAV the mechanism of PKP2-mediated viral restriction. First, polymerase activity. Indeed, PKP2 competed with PB2 and we examined the interaction of PB1 with endogenous PKP2. reduced B70% of PB2 protein in the PB1 complex (Fig. 6a), Primary human tracheal epithelial cells were infected with PR8 whereas the PB1–PA and PB1–NP interactions remained intact IAV and then lysed for immunoprecipitation. Following PR8 IAV (Fig. 6b,c). We also found that NP, PB2 and PA failed to pull infection, endogenous PKP2 bound to viral PB1 (Fig. 3a). Next, down PKP2, suggesting the interaction specificity between PB1 we examined PKP2-PB1 co-localisation. A549 lung epithelial cells and PKP2 (Supplementary Fig. 10a). To investigate the effect of were infected with PR8 IAV. Consistent with a previous report9, PKP2 on the IAV polymerase complex, we first infected HEK293 endogenous PKP2 was expressed in cell junctions and nuclei. cells stably expressing PKP2-FLAG or GFP-FLAG with 1 MOI of Following infection, PB1 was expressed and co-localized with PR8 IAV for 16 h. Then, we fractionated the cell lysates by PKP2 in the nucleus (Fig. 3b). Finally, we determined which centrifugation over a 15–55% sucrose gradient. Notably, PKP2 domains were required for the PB1 and PKP2 interaction. The co-fractionated with the majority of IAV polymerase complex PB1 C-terminal region consisting of residues 493 to 757 was distributed from fractions #4 to #7, whereas green fluorescent sufficient for the interaction with PKP2 (Fig. 3c), whereas the protein was absent in these fractions (Fig. 6d). Furthermore, N-terminal domain of PKP2 was required for the interaction with PKP2 perturbed the fraction distribution of IAV polymerase PB1 (Fig. 3d). complex and reduced the PB2/PB1 protein ratio in fractions

4 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ARTICLE

a

PB2 PB1

M2 NS1 NP

Bait New interaction

HCIP Known interaction

bc4.5 A549 Primary tracheal cells * 2.5 4 * HCIP EIF2B4 3.5 2 3 * 2.5 * * 1.5 +2SD * 2 * * 1 1.5

1 0.5 –2SD 0.5 * * TRIM41 Relative PR8-Gluc reporter activity FKBP8 Relative PR8-Gluc reporter activity 0 PKP2 ZMPSTE24 0 RNAi: PKP2 Control EIF2B4 EKBP8 TRIM41 ZMPSTE24 Figure 2 | Identifying HCIPs that regulate IAV infection by screening the core interactome. (a) Map of the core IAV–host protein interactome that comprises HCIPs associated with multiple IAV strains in Fig. 1b. Legends are indicated. (b) PR8-Gluc reporter screening assay in HEK293 cells. At 24 h after transfection with HCIPs, the cells were infected with 0.1 MOI of PR8-Gluc. After 16 h, IAV infection activity was determined by luciferase activity. The screenings were biologically repeated two times. The dash lines indicate ±2 X s.d. of the whole screening data set. The HCIPs with 42 Â s.d. or o À 2 Â s.d. are labelled. (c) A549 and primary human tracheal epithelial cells were transfected with control siRNA or the indicated siRNA duplexes. After 48 h, the cells were infected with 0.1 MOI of PR8-Gluc for 16 h. The relative luciferase activity was examined. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the corresponding siRNA control in each cell group. An asterisk indicates Po0.05.

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876

ab Anti- PB1 PKP2 DAPI Merge IP: PKP2

Control IgG IAV PR8 IAV – ++ MW 0 h 75

WB: anti-PB1 20 μm

75 Input 5% WB: anti-PKP2 8 h 75

WB: anti-PB1

c d 837 PA binding Polymerase domain PB2 binding Armadillo repeat 9 PB1 1 757 8 7 6 17 266 493 678 PKP2 1 1 2 345 N266 1 266 348 574 635 837 N348 1 348 9 PD 267 493 8 7 6 C493 493 757 C349 349 1 2 345

PKP2-V5 PB1-HA PKP2 Vector N348 C349 Vector PB1 N266 PD C493

75 75 IP: anti-FLAG WB: anti-V5 IP: anti-FLAG WB: anti-HA

75 75

50 37 37 15 IP: anti:FLAG WB: anti-FLAG IP: anti-FLAG WB: anti-FLAG

75 75 Lysate WB: anti-HA Lysate WB: anti-VA

Figure 3 | PKP2 interacts and co-localizes with PB1. (a) Primary human tracheal epithelial cells were infected with 1 MOI of PR8 IAV for 16 h. Cell lysates were subjected to immunoprecipitation and immunoblotting with the indicated antibodies to detect endogenous interactions. Molecular weights (MW)are indicated. (b) A549 cells were infected with 1 MOI of PR8 IAV for 8 h and stained with anti-PKP2 (green), anti-PB1 (red) and DAPI nuclear stain (blue). (c) Full-length and various PB1 deletion mutants were fused with FLAG epitope and co-transfected with PKP2-V5 into HEK293 cells. Lysates were immunoprecipitated with anti-FLAG antibody and blotted with the indicated antibodies. (d) Full-length and various PKP2 deletion mutants were fused with FLAG epitope and co-transfected with PB1-HA into HEK293 cells. Lysates were immunoprecipitated with anti-FLAG antibody and blotted with the indicated antibodies. containing PKP2 (Fig. 6d). Finally, we examined whether and inhibits the cleavage and polyadenylation of host PKP2 had an effect on PB1 polymerase activity using a pre-mRNA25. In our network, CPSF, FIP1L1 and WDR33 polymerase reporter assay. Silencing PKP2 with siRNA strongly interact with NS1 of WSN/33 and Aichi, but not with enhanced polymerase activity, whereas overexpression of PKP2 PR8 NS1. NS1 of PR8 influenza lacks F103 and M106 residues reduced polymerase activity (Fig. 6e and Supplementary Fig. 10b). that are required to stabilize the CPSF interaction21. Notably, NS1 Thus, PKP2 alters the integrity of IAV polymerase complex and of NY/2009 weakly interacts with CPSF even though it possesses impairs IAV replication. intact F103 and M106. Dr Garcia-Sastre’s group also suggest that other binding sites are also required26. In addition to CPSF, many Discussion new strain-specific NS1 interactors were identified, including We constructed and comparatively analysed several IAV–host FKBP5, NKRF, TNFAIP8, ZFC3H1 and TRNAU1AP. FKBP5, protein interactomes. As with any screening approach, the NKRF and TNFAIP8 are involved in NF-kB signalling27–29. database does not represent a complete interaction network and ZFC3H1 was recently shown to be required for IL-8 many of the hits identified in this screen require additional production30. These unique interactions suggest distinct validation using more directed experiments. Nonetheless, the perturbation strategies of each IAV strain, potentially enabling comparative interactomes provide insights on the virulence and the definition of de novo virulence and host determinants. host determinants. For example, the various NS1 proteins exhibit Systematic comparisons of strain-specific influenza–host different binding abilities to CPSF proteins and the CPSF-binding protein interactomes are also essential for the identification of partners FIP1L1 and WDR33 (ref. 24). NS1 interacts with CPSF general mechanisms of regulating viral infection and the

6 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ARTICLE

a b c 1.2 Vector 1.2 PKP2 Vector PKP2-FLAG 1 –2 –1 –2 –1 1 * WSN (MOI) 0 10 10 101010 1 0.8 50 0.8 0.6 WB: anti-NP 0.4 0.6 75 0.2 0.4 WB: anti-FLAG 50 0 0.2 Relative PR8-Gluc reporter activity WB: anti-β- Relative viral infection activity PKP2 PKP4

Vector 0

CTNNB1 IAV VSV SeV ef 75 Vector Vector WB: anti-FLAG 1.2 PKP2 107 PKP2

**** )

1 –1 * * d 106 Vector PKP2 0.8 * 0.6 105 NP 0.4 DAPI 4

Relative IAV infection 0.2 10 μ 100 M IAV WSN/33 titer (pfu ml 0 103 PR8 Aichi h.p.i. 24 h 48 h 72 h WSN/33 NY/2009

Figure 4 | PKP2 restricts IAV infection. (a) HEK293 cells were transfected with pCMV3-tag-8 vector, PKP2-FLAG, PKP4-FLAG or CTNNB1 (b-catenin)- FLAG for 24 h, followed by infection with 0.1 MOI of PR8-Gluc for 16 h. Luciferase activity assays were performed. All experiments were biologically repeated three times. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the vector control. An asterisk indicates Po0.05. Western blot shows the protein expression levels from one experiment. (b) HEK293 cells transfected with PKP2-FLAG or pCMV3-tag-8 vector were infected with 0.1 MOI of IAV-Gluc, VSV-Luc and SeV-Luc for 16 h. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the vector control. An asterisk indicates Po0.05. (c) HEK293 cells were transfected with pCMV3-tag-8 vector or PKP2-FLAG for 24 h, then infected with WSN/33 IAV for 16 h. Cell lysates were blotted using the indicated antibodies. (d) HEK293 cells were transfected with pCMV3-tag-8 vector or PKP2-FLAG for 24 h, then infected with WSN/33 IAV for 16 h. Fixed cells were stained with an anti-NP antibody. The percentage of stained cells is summarized in e.(e) HEK293 cells were transfected with the pCMV3-tag-8 vector or PKP2-FLAG for 24 h, then infected with 1 MOI of PR8, WSN/33, NY/2009 or Aichi IAV for 16 h. Fixed cells were stained with anti-NP antibody (PR8 and WSN/33), anti-HA (NY/2009) or anti-Aichi antibodies. The relative infection was determined by the ratio of positive cells. Data represent means ±s.d. of three independent experiments (480 cells counted per experiment). The P value was calculated (two-tailed Student’s t-test) by comparison with the pCMV3-tag-8 vector control. An asterisk indicates Po0.05. (f) A549 cells were transfected with PKP2-FLAG. After 48 h, cells were infected with 0.001 MOI of WSN/33 IAV. After the designated hour post infection (h.p.i.), virus titers were determined by plaque assay. All experiments were biologically repeated three times. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the pCMV3-tag-8 vector control. An asterisk indicates Po 0.05. discovery of common therapeutic targets. For example, IPO5 is a PKP2 interacted with PB1, suggesting the interaction specificity major common interactor of PB1 proteins and required for IAV between PKP2 and PB1. Consistently, PKP2, but not other PKPs polymerase nuclear import19. The heat shock protein Hsp90 is a restrict IAV infection. host factor facilitating viral RNA synthesis31. In our IAV–host PKPs are scaffold proteins that are essential for the formation interactomes, the heat shock proteins DNAJCs are common of desmosome and stabilisation of cell junctions. PKP2 is the interactors for PB1 and PB2. Whether these heat shock proteins largest and most prevailing plakophilin protein that is expressed regulate viral RNA synthesis requires further investigation. in all cell types with desmosomal junctions9. Although PKP2 was By functional screening the core interactome, we identified four originally identified as cell junction protein anchoring on plasma common interactors (ZMPSTE24, FKBP8, PKP2 and TRIM41) membrane, it also localizes in the nucleus9. Interestingly, nuclear that restrict IAV infection and may lead to the discovery of new PKP2 associates with the largest subunit of the RNA polymerase general antiviral mechanisms. III holoenzyme subunit of transcription factor IIIB (TFIIIB), an In the core interactome, PKP2 interacts with influenza PB1 enzyme important for transcription of ribosomal RNA and proteins and restricts IAV infection. PKP2, along with PKP1 and tRNA35. However, the biological significance of the interaction PKP3 form a subgroup of catenin protein family that is with RNA polymerase III is not known. characterized by the armadillo repeats. PKPs are composed of a Our data demonstrate that PKP2 binds to the C-terminal basic N-terminal head domain, followed by nine armadillo domain of PB1, thus blocking the interaction of PB2 with PB1. repeats and a short C-terminal tail32,33. A spacer between the fifth Assembly of the heterotrimeric IAV polymerase requires and sixth repeat results in a characteristic kink in the domain interactions between the PB1 N-terminal domain with the PA structure34. The N terminus of PKPs is diverse and exhibits no chain, whereas the PB1 C-terminus associates with the poly- obvious homology. Our study showed that the N terminus of merase PB2 protein36–38. The interactions between the

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876

ab 4 * *

3 IAV PR8-Gluc 3 * PKP2 mRNA 2

2 * 1 Relative PR8-Gluc activity Relative 0 1 Relative levels Relative Control RNAi + –– PKP2 RNAi – + + * PKP2-FLAG – – + * 0 RNAi: Control #1 #2 #3 PKP2-FLAG 75 Endo-PKP2 PKP2 WB: anti-PKP2 50

WB: anti-β-actin

Control RNAi cdPKP2 RNAi 4 * Control RNAi 8 3.5 * * * 10 PKP2 RNAi * 3 ) * –1 7 2.5 10 2 * 1.5 106 1 Relative IAV infection IAV Relative 0.5 105

0 WSN/33 titer (pfu ml IAV

4 PR8 Aichi 10 NY/2009 WSN/33 h.p.i. 24 h 48 h 72 h

Figure 5 | RNAi depletion of PKP2 promotes influenza virus infection. (a) A549 cells were transfected with scrambled control siRNA and 3 siRNA oligos against PKP2. After 48 h, the cells were infected with 0.1 MOI of IAV PR8-Gluc for 16 h. Relative luciferase activities were examined and PKP2 mRNA levels were determined by RT-PCR. All experiments were biologically repeated three times. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the siRNA control in each cell group. An asterisk indicates Po0.05. (b) HEK293 cells were transfected with siRNA targeting the 3’-UTR of PKP2 (duplex #1) and/or PKP2-FLAG resistant to PKP2 siRNA. After 48 h, the cells were infected with 0.1 MOI of PR8-Gluc for 16 h. The relative luciferase activity is shown. All experiments were biologically repeated three times. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the siRNA control. An asterisk indicates Po0.05. Western blot demonstrates the knockdown efficiency. FLAG-tagged PKP2 and endogenous PKP2 (Endo-PKP2) are indicated. (c) A549 cells were transfected with PKP2 siRNA (duplex #1) for 48 h, then infected with 1 MOI of PR8, WSN/33, NY/2009 or Aichi IAV. Fixed cells were stained with anti-NP antibody (PR8 and WSN/33), anti-HA (NY/2009) or anti-Aichi antibodies. The relative infection was determined from the ratio of positive cells. Data represent means ±s.d. of three independent experiments (480 cells counted per experiment). *Po0.05 significantly different versus control RNAi as calculated by the two-tailed Student’s t-test. (d) Tracheal epithelial cells transfected with control or PKP2 siRNA were infected with 0.001 MOI of WSN/33 IAV for the indicated h.p.i. Titers of culture supernatants containing IAV were determined on MDCK cells, and plaques were enumerated. Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the control RNAi. An asterisk indicates Po0.05. polymerase subunits are attractive drug targets because proper Methods assembly of these subunits is crucial for normal polymerase Antibodies. Anti-b-actin (Abcam, # ab8227, WB (1:1,000)), anti-FLAG (Sigma, # F3165, WB (1:1,000), IFA (1:100)), anti-GFP (Santa Cruz Biotechnology, activity. Several inhibitors of synthetic peptides based on the # sc-9996, WB (1:2,000)), anti-HA epitope (Sigma, # H3663, WB (1:1,000)), PB1–PB2 interaction domain have been reported to interfere with anti-NP (BEI resources, # NR-4282, WB (1:1,000), IFA (1:100)), anti-NP (EMD PB1–PB2 interactions and disrupt normal function of the Millipore, # MAB8800, IFA (1:1,000)), anti-HA from A/California/04/2009 (H1N1) polymerase39,40. Our study suggests that PKP2 is a natural (BEI resources, # NR-42021, IFA (1:100)), anti-influenza A/Aichi/2/68 (BEI resources, # NR-3125, IFA (1:100)), anti-PB1 (BEI resources, # NR-31690, WB (1:1,000)), inhibitor of IAV polymerase complex. Defining the critical anti-PB2 (Genetex, # GTX125926, WB (1:1,000)), anti-PA (Genetex, # GTX125932, interaction sites of PKP2 will provide a basis for the development WB (1:1,000)), anti-PKP2 (Fitzgerald Industries International, # 20R-2656, of new antivirals. WB (1:1,000), IFA (1:100)), anti-CPSF4 (Millipore, # MABE620, WB (1:1,000)). In summary, these comparative IAV–host interactomes will Goat anti-Mouse IgG-HRP (Santa Cruz Biotechnology, # sc-2055, WB (1:10,000)), Goat anti-Rabbit IgG-HRP (Santa Cruz Biotechnology, # sc-2030, WB contribute to the understanding of host defence mechanisms and (1:10,000)), Goat anti-Guinea Pig IgG-HRP (Thermo Fisher Scientific, # A18775, might ultimately provide the insights and opportunities necessary WB (1:10,000)), Alexa Fluor 594 Goat Anti-Mouse IgG (H þ L) (Life Technologies, for the identification of new antiviral drug targets. # A11005, IFA (1:200)), Alexa Fluor 488 Goat Anti-Rabbit IgG (H þ L)

8 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ARTICLE

a b c PKP2-V5+ – + – + PKP2-V5 +–+–+ PKP2-V5 +–+–+ PB2-HA – + – + + NP –+– ++ PA – +–+ + PB1-FLAG ––+++ PB1-FLAG ––+++ PB1-FLAG ––+++ 1.0 0.3 50 75 75 Anti-NP Anti-PA Anti-HA 75 75 75 Anti-V5 Anti-V5

Anti-V5 anti-FLAG IP: IP: anti-FLAG IP: 75 IP: anti-FLAG IP: 75 75 Anti-FLAG Anti-FLAG Anti-FLAG 75 50 Anti-PA 75 Anti-NP Input

Anti-HA Input 75 Input 75 75 Anti-V5 Anti-V5 Anti-V5

d PKP2-FLAG GFP-FLAG

Fraction # 123456789101112131412345 6 7 8 9 1011121314 PB2/PB1 1.0 1.5 1.9 1.6 3.2 2.9 3.2 3.1

75 75 WB: anti-PB2

75 75 WB: anti-PB1

50 50 WB: anti-NP

75 75 WB: anti-PA

75 25 WB: anti-FLAG e 1.2 4 * * 1 3 0.8

0.6 2

0.4 1 0.2

Relative IAV polymerase activity polymerase IAV Relative 0 0 Vector PKP2 Control PKP2 RNAi RNAi

Figure 6 | PKP2 perturbs the IAV polymerase complex and inhibits viral polymerase activity. (a) PB1 and PB2 were co-transfected with PKP2 into HEK293 cells. After 48 h, cell lysates were harvested for immunoprecipitation using anti-FLAG antibody and blotted using the indicated antibodies. (b) PB1 and NP were co-transfected along with PKP2 into HEK293 cells. After 48 h, cell lysates were harvested for immunoprecipitation using anti-FLAG antibody. The indicated antibodies were used for blotting. (c). PB1 and PA were co-transfected with PKP2 into HEK293 cells. After 48 h, cell lysates were harvested for immunoprecipitation using anti-FLAG antibody and blotted as indicated. (d) HEK293 cells stably expressing PKP2-FLAG or GFP-FLAG were infected with 1 MOI of PR8 IAV for 16 h. Then the cell lysates were separated by 15–55% sucrose density centrifugation. Fractions were blotted using the indicated antibodies. The ratios of PB2 to PB1 in the fractions 4–7 were indicated. (e) HEK293 cells were transiently transfected with a plasmid cocktail containing PB1, PB2, PA, NP expression plasmids of PR8 IAV plus a polymerase I plasmid expressing an influenza virus-like RNA encoding the reporter protein firefly luciferase, along with control siRNA, PKP2 siRNA, the vector pCMV3-tag-8 or PKP2-FLAG for 48 h. The relative luciferase signal is shown. The transfection efficiency was determined by western blot (Supplementary Fig. 10b). Data represent means ±s.d. of three independent experiments. The P value was calculated (two-tailed Student’s t-test) by comparison with the corresponding control. An asterisk indicates Po0.05.

(Life Technologies, # A11034, IFA (1:200)), Alexa Fluor 488 Goat Anti-Guinea Pig Webster (St Jude Children’s Hospital, Memphis, TN)42. NY/2009 IAV viral genes IgG (H þ L) (Life Technologies, # A11073, IFA (1:200)). are generous gift from J. Craig Venter Institute (La Jolla, CA). NS1 of VN/2004 IAV is a kind gift from Dr Andrew Rice (Baylor College of Medicine, TX). The NP of Aichi IAV (NCBI, # AFM71861.1) was ordered from Sino Biological Inc., Plasmids. PR8 IAV viral genes are kind gift from Dr Ervin Fodor (University of Beijing, China (# VG40207-UT). The PB1, PB2 and PA of VN/2004 IAV are Oxford, Oxford, UK)41. WSN/33 IAV viral genes are generous gift from Dr Robert generous gift from Koyu Hara (Kurume University School of Medicine, Japan)43.

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876

The M2 of VN/2004 IAV (NCBI, # EF541452) and NS1 of Aichi IAV (NCBI, Sciences, Marlborough, MA). Uncropped scans of western blots are provided in # AFM71862.1) were synthesized by GenScript (Piscataway, NJ). Various PB1 Supplementary Fig. 11. mutants of IAV PR8 were constructed using a QuikChange II Site-Directed For immunoprecipitation, 2% of cell lysates were saved as an input control, and Mutagenesis kit (Agilent Technologies, # 200523). the remainder was incubated with 5–10 ml of the indicated antibody plus 20 mlof Human PKP2-FLAG was a gift from Kathleen Green (Addgene plasmid Pierce Protein A/G Plus Agarose (Thermo Fisher Scientific, # 20423), 10 mlof # 32230)44. Point mutations and deletions of PKP2-FLAG were constructed using a EZview Red Anti-FLAG M2 Affinity Gel (Sigma, # F2426), or 10 ml of Anti-HA- QuikChange II Site-Directed Mutagenesis kit. Human PKP4-FLAG (a.k.a Agarose (Sigma, # A2095). After mixing end-over-end at 4 °C overnight, the beads p0071-FLAG) was a gift from Dr Andrew Kowalczyk (Emory University, Atlanta, were washed 3 times (1 min each wash) with 500 ml of lysis buffer. Anti-HA À 1 GA)45. Human EIF2B4-V5 was purchased from Harvard PlasmID Database Agarose and Anti-FLAG M2 Affinity Gel were eluted with 50 ml of 1 mg ml À 1 (# HsCD00414983). Human TRIM41-HA was a generous gift from Dr Adolfo HA peptide (Sigma, # I1249) or 5 mg ml 3X FLAG peptide (Sigma, # F4799), Garcia-Sastre (Mount Sinai School of Medicine, NY)46. Human ZMPSTE24 cDNA respectively. Immunoprecipitated complexes with Protein A/G Plus Agarose were (Harvard PlasmID Database, # HsCD00075979) was cloned into pCMV-3Tag-8 mixed with 1 Â Lane Marker Reducing Sample Buffer. (Agilent Technologies, # 240203) to make ZMPSTE24-FLAG. Human FKBP8 cNDA (Harvard PlasmID Database, # HsCD00082457) was cloned into pCMV-3Tag-8 to make FKBP8-FLAG. CTNNB1-FLAG was from Dr He Immunofluorescence assay. Cells were cultured in the Lab-Tek II CC2 Chamber laboratory as described47. The Wnt signalling reporter, TOPFLASH was purchased Slide System 4-well (Thermo Fisher Scientific, # 154917). After the indicated from Upstate Biotechnology (#21–170). The control plasmid, pCMV-3Tag-8 was treatment, the cells were fixed and permeabilized in cold methanol for 10 min at purchased from Agilent Technologies (#240203). À 20 °C. Then, the slides were washed with 1 Â PBS for 10 min and blocked with Odyssey Blocking Buffer (LI-COR Biosciences, # 927-40000) for 1 h. The slides were incubated in Odyssey Blocking Buffer with appropriately diluted primary ° Cells. HEK293 cells (ATCC, # CRL-1573) and MDCK cells (ATCC, # CCL-34) antibodies at 4 C for 16 h. After 3 washes (5 min per wash) with 1 Â PBS, the cells were maintained in Dulbecco’s Modified Eagle Medium (Life Technologies, were incubated with the corresponding Alexa Fluor conjugated secondary anti- # 11995-065) containing antibiotics (Life Technologies, # 15140-122) and 10% fetal bodies (Life Technologies) for 1 h at room temperature. The slides were washed bovine serum (Life Technologies, # 26140-079). A549 cells (ATCC, # CCL-185) three times (5 min each time) with 1 Â PBS and counterstained with 300 nM DAPI were cultured in RPMI Medium 1640 (Life Technologies, # 11875-093) plus 10% for 1 min, followed by washing with 1 Â PBS for 1 min. After air-drying, the slides fetal bovine serum and 1 Â MEM Non-Essential Amino Acids Solution were sealed with Gold Seal Cover Glass (Electron Microscopy Sciences, # 3223) (Life Technologies, # 11140-050). Primary human tracheal epithelial cells and using Fluorogel (Electron Microscopy Sciences, # 17985-10). Images were captured supporting medium were purchased from Lifeline Technology (Frederick, MD) and analysed using an Olympus Fluoview Confocal microscope (Olympus, Center (# FC-0035 and # LL-0023). The cell lines were examined using MycoAlert Valley, PA). Mycoplasma Detection Kit (Lonza, # LT07-418). Real-time PCR. Total RNA was prepared using RNeasy columns (Qiagen, # 74106). In total, 1 mg quantity of RNA was transcribed into cDNA using a Viruses . Influenza A/Puerto Rico/8/34 (H1N1) (Charles River Laboratories, QuantiTect reverse transcription kit (Qiagen, # 205311). For one real-time reaction, Wilmington, MA, # 10100374), influenza A/WSN/33 (H1N1) (a kind gift of 20 ml of SYBR Green PCR reaction mix (Roche Applied Science) including a 1/10 Dr Peter Palese, Mount Sinai School of Medicine, NY), influenza A virus A/New volume of the synthesized cDNA plus an appropriate oligonucleotide primer pair York/18/2009 (H1N1) pdm09 (BEI Resources, # NR-15268), and A/Aichi/68 were analysed on a LightCycler 480 (Roche). The comparative Ct method was used (H3N2) (Charles River Laboratories, # 10100375). Influenza PR8-GLuc virus was a to determine the relative mRNA expression of genes normalized by the house- generous gift from Dr Peter Palese and features a Gaussia luciferase (Gluc) gene keeping gene GAPDH. The primer sequences: PKP2, forward primer 50-GTGGG inserted downstream of PB2 (ref. 48). IAV was propagated in specific pathogen- CAACGGAAATCTTCAC-30, reverse primer 50-CCAGCCTTTAGCATGTCAT free fertilized eggs Premium Plus (Charles River Laboratories) as described by 0 0 0 49 AGG-3 ; GAPDH, forward primer 5 -AGGTGAAGGTCGGAGTCA-3 , reverse Szretter et al. In total, 9–11-day-old embryonated chicken eggs are used for the primer 50-GGTCATTGATGGCAACAA-30. production of influenza virus. 0.2 ml stock influenza virus at 1 Â 103 TCID50 was injected through the puncture hole into the allantoic cavity. After 72 h of incubation, allantoic fluid was collected. IAV titers were determined by plaque Plasmid transfection 50 6 . HEK293 cells were transfected using Lipofectamine 2000 assay as described by Matrosovich et al. In brief, 1.2 Â 10 MDCK cells per ml Transfection Reagent, and A549 cells were transfected using Lipofectamine were split into six-well plates. After 2 Â washes with Dulbecco’s Modified Eagle 3000 Transfection Reagent (Life Technologies, # L3000015) according to the Medium, serial dilutions of IAV were adsorbed onto the cells for 1 h. The cells were manufacturer’s protocol. In total, 0.5 mg HCIP plasmid was used for overexpression covered with Dulbecco’s Modified Eagle Medium containing 1 Â Avicel RC591 NF experiments. (FMC Biopolymer, Philadelphia, PA) and 1 mgmlÀ 1 TPCK-trypsin (Thermo Fisher Scientific, # 20233). Crystal violet staining was performed 48 hour post infection , and visible plaques were counted. RNAi depletion. RNAi target sequences (sense strand): PKP2 siRNA #1: 50- ACG Sendai virus with luciferase was a kind gift from Dr Charles Russell (St. Jude’s GCTCATGTTAATGAGTTA-30; PKP2 siRNA #2: 50- TCCGTGGGCAACGGAAA Hospital, Memphis, TN). Vesicular stomatitis virus with firefly luciferase gene TCTT-30; PKP2 siRNA #3: 50-ATCCAGCGAAATGAATCTACA-30. ZMPSTE24 (VSV-Luc) was a kind gift from Dr Sean Whelan (Harvard Medical School, siRNA: 50-CAATCTATGCTGATTATAT-30. EIF2B4 FlexiTube siRNA (Qiagen, Boston, MA). # GS8890). TRIM41 FlexiTube siRNA (Qiagen, # GS90933). FKBP8 FlexiTube siRNA (Qiagen, # GS23770). siGENOME Non-Targeting Control siRNA (Dharmacon, # D-001210-02-05) was used as the control siRNA. siRNA duplexes Sample preparation, western blot and immunoprecipitation. Approximately were transfected into A549 (5 pmol) and primary tracheal cells (5 pmol) using 1 Â 106 cells were lysed in 500 ml of tandem affinity purification lysis buffer (50 mM Lipofectamine RNAiMAX Transfection Reagent (Life Technologies, # 13778030) Tris-HCl (pH 7.5), 10 mM MgCl2, 100 mM NaCl, 0.5% Nonidet P40, 10% glycerol, according to the manufacturer’s protocol. Complete EDTA-free protease inhibitor cocktail tablets (Roche, # 11873580001)) for 30 min at 4 °C. The lysates were then centrifuged for 30 min at 15,000g. Supernatants were collected and mixed with 1 Â Lane Marker Reducing Sample Sucrose gradient centrifugation. A total of 1 Â 108 HEK293 cells stably Buffer (Thermo Fisher Scientific, # 39000). expressing FLAG-tagged PKP2 were collected and lysed in gradient centrifugation Precision Plus Protein Dual Color Standards (5 ml) (Bio-Rad, # 161-0374) (GC) buffer (25 mM Tris Á HCl (pH 7.5), 150 mM KCl, 0.5% Nonidet P40, 2 mM and samples (10–15 ml) were loaded into Mini-Protean TGX Precast Gels, 15 well EDTA, 1 mM Sodium fluoride, 0.5 mM dithiothreitol, phosphatase inhibitors, and (Bio-Rad, # 456-103), and run in 1 Â Tris/Glycine/SDS Buffer (Bio-Rad, protease inhibitors). The lysates were centrifuged at 16,000g for 30 min at 4 °Cto # 161-0732) for 35 min at 200 V. Protein samples were transferred to Immun-Blot remove cell debris. The lysates were then separated in GC buffer with 15% PVDF Membranes (Bio-Rad, # 162-0177) in 1 Â Tris/Glycine Buffer (Bio-Rad, (wt vol À 1) to 55% (wt vol À 1) sucrose by ultracentrifugation in an SW41 Ti rotor # 161-0734) at 70 V for 60 min. PVDF membranes were blocked in 1 Â TBS buffer (Beckman Coulter, Inc.) at 111,000g for 18 h at 4 °C. Fractions were collected and (Bio-Rad, # 170-6435) containing 5% Blotting-Grade Blocker (Bio-Rad, loaded onto SDS-PAGE gels for western blotting. # 170-6404) for 1 h. After washing with 1 Â TBS buffer for 30 min, the membrane blot was incubated with appropriately diluted primary antibody in antibody dilution buffer (1 Â TBS, 5% BSA, 0.05% sodium azide) at 4 °C for 16 h. Then, the Reconstitution of influenza virus polymerase activity. HEK293 cells were blot was washed three times with 1 Â TBS (each time for 10 min) and incubated transfected with vectors expressing PR8 PB1(0.05 mg), PB2 (0.05 mg), NP (0.05 mg), with secondary HRP-conjugated antibody in antibody dilution buffer (1:10000 PA (0.05 mg) and the indicated PKP2 (0.2 mg) or siPKP2 duplex (5 pmol) in dilution) at room temperature for 1 h. After three washes with 1 Â TBS (each time addition to a polymerase I (PolI)-driven plasmid transcribing an IAV-like for 10 min), the blot was incubated with Clarity Western ECL Substrate (Bio-Rad, RNA encoding the reporter protein firefly luciferase (0.1 mg) to monitor viral # 170-5060) for 1–2 min. The membrane was removed from the substrates, then polymerase activity40. Cells were lysed 48 h after transfection. Luciferase activity exposed either to HyBlot CL Autoradiography Film (Denville Scientific Inc. was measured with a luciferase assay system (Promega, Madison, WI). Transfection # E3018) in the dark room or to Amersham imager 600 (GE Healthcare Life efficiency was monitored by western blot.

10 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876 ARTICLE

Cell viability. Cell viability was assessed by using the CellTiter-Glo Luminescent score. The fold change was calculated for each prey protein as the ratio of spectral Cell Viability Assay (Promega, # G7570) according to the manufacturer’s counts from replicate bait purifications to the spectral counts of the same prey instructions. protein across all negative controls. A background factor of 0.1 was added to the average spectral counts of negative controls to prevent division by zero. The proteins included in the final interactome list had an AvgPZ0.89. The threshold Stable cell line selection . Two mirograms of each viral construct was transfected for SAINT scores was selected based on receiver operating curve analysis into HEK293 cells using Lipofectamine 2000 (Invitrogen, # 11668027). Two days performed using publicly available protein interaction data. A SAINT score of m À 1 m À 1 after transfection cells were treated with 3 gml puromycin or 100 gml AvgPZ0.89 was considered a true positive BioID protein with an estimated FDR of hydromycin for 14 days. Single colonies were picked and expanded in six-well r2%. Proteins with SAINT scoreo0.89 are considered as non-specific binding plates. Protein expression levels in each colony were determined by proteins. We manually removed ribosomal proteins from the final HCIP list immunoblotting. because these proteins are prone to associate with RNA-binding proteins. The results reported here will require additional validation using more directed Purification of protein complexes and mass spectrometry. AP-MS experiments experiments. Low affinity or transient, and may be lost during the stringent were performed as previously described12. For protein purification, HEK293 cell washing procedures with lysis buffer. Although spectral counts have a correlation lines stably expressing each FLAG-tagged viral protein were divided into two with protein abundance other methods, for example, SILAC offer more precise groups. Then, one group was infected with 1 MOI of IAV. After 16 h of infection, strategies for quantitative proteomics53. Nevertheless, by mapping the IAV–host 108 cells were collected and lysed in 10 ml of tandem affinity purification buffer4. interactomes using AP-MS, we were able to reveal novel common and strain- Cell lysates were precleared with 50 ml of protein A/G resin before the addition of specific interactors. 20 ml of anti-FLAG resin (Sigma, # F2426) and incubation for 16 h at 4 °Cona rotator. The resin was washed three times and transferred to a spin column with 40 ml of 3X FLAG peptide for 1 h at 4 °C on a rotator. The purified complexes were Statistical analysis. The sample size was sufficient for data analysis using paired loaded onto a 4–15% NuPAGE gel. The gels were stained with a SilverQuest two-tailed Student’s t-test. For all statistical analysis, differences were considered to staining kit (Invitrogen), and lanes were excised for mass spectrometry analysis be statistically significant at values of Po0.05. by the Taplin Biological Mass Spectrometry Facility (Harvard Medical School, Boston, MA). Bioinformatic analysis. The Search Tool for the Retrieval of Interacting B 3 Genes/Proteins (STRING) database (http://string.embl.de) and curated Mass spectrometry. Excised gel bands were cut into 1mm pieces. Gel pieces literature5–8,17,54,55 were used to identify novel HCIPs. The IAV–host protein were then subjected to a modified in-gel trypsin digestion procedure. Gel pieces interaction network was generated in Cytoscape (www.cytoscape.org). DAVID were washed and dehydrated with acetonitrile for 10 min. followed by removal of Bioinformatics Resources 6.8 (ref. 56) was used to perform acetonitrile. Pieces were then completely dried in a speed-vac. Gel pieces were À 1 enrichment analysis of the 625 HCIPs from comparative IAV–host protein rehydrated with 50 mM ammonium bicarbonate solution containing 12.5 ng ml interactomes. Terms within the top 5 ‘GOTERM_BP_DIRECT’ category were used. modified sequencing-grade trypsin (Promega, Madison, WI) at 4 °C. After 45 min., the excess trypsin solution was removed and replaced with 50 mM ammonium bicarbonate solution to just cover the gel pieces. Peptides were later extracted by Data availability. The authors declare that the data supporting the findings removing the ammonium bicarbonate solution, followed by one wash with a of this study are available within the article and its Supplementary solution containing 50% acetonitrile and 1% formic acid. The extracts were then B Information files, and from the corresponding author upon request. The AP-MS dried in a speed-vac ( 1 h) and stored at 4 °C until analysis. data that support the findings of this study have been deposited in the IMEx On the day of analysis the samples were reconstituted in 5–10 ml of HPLC (http://www.imexconsortium.org) consortium through IntAct57 with the accession solvent A (2.5% acetonitrile, 0.1% formic acid). A nano-scale reverse-phase HPLC code IM-25584 and also can be downloaded from https://sharehost.hms. capillary column was created by packing 5 mm C18 spherical silica beads into a harvard.edu/mbib/dorf/nature_communications. fused silica capillary (100 mm inner diameter  B12 cm length) with a flame- drawn tip. After equilibrating the column each sample was loaded via a Famos auto sampler (LC Packings, San Francisco CA) onto the column. A gradient was formed References and peptides were eluted with increasing concentrations of solvent B (97.5% 1. Tscherne, D. M. & Garcia-Sastre, A. Virulence determinants of pandemic acetonitrile, 0.1% formic acid). As peptides eluted they were subjected to electrospray ionisation and then influenza viruses. J. Clin. Invest. 121, 6–13 (2011). entered into an LTQ Velos ion-trap mass spectrometer (Thermo Fisher, San Jose, 2. Fukuyama, S. & Kawaoka, Y. The pathogenesis of influenza virus infections: the CA). Peptides were detected, isolated and fragmented to produce a tandem mass contributions of virus and host factors. Curr. Opin. Immunol. 23, 481–486 spectrum of specific fragment ions for each peptide. Dynamic exclusion was (2011). enabled such that ions were excluded from reanalysis for 30 s. Peptide sequences 3. Takeuchi, O. & Akira, S. Pattern recognition receptors and inflammation. Cell (and hence protein identity) were determined by matching protein databases with 140, 805–820 (2010). the acquired fragmentation pattern by the software program, Sequest (Thermo 4. Fu, B. et al. TRIM32 senses and restricts influenza A virus by ubiquitination of Fisher, San Jose, CA). The human IPI database (Ver. 3.6) was used for searching. PB1 polymerase. PLoS Pathog. 11, e1004960 (2015). Precursor mass tolerance was set to ±2.0 Da and MS/MS tolerance was set to 5. Tripathi, S. et al. Meta- and orthogonal integration of influenza ‘OMICs’ data 1.0 Da. A reversed-sequence database was used to set the false discovery rate at 1%. defines a role for UBR4 in virus budding. Cell Host Microbe 18, 723–735 Filtering was performed using the Sequest primary score, Xcorr and delta-Corr. (2015). Spectral matches were further manually examined and multiple identified peptides 6. Watanabe, T. et al. influenza virus-host interactome screen as a platform for (4 ¼ 2) per protein were required. antiviral drug development. Cell Host Microbe 16, 795–805 (2014). 7. Heaton, N. S. et al. targeting viral proteostasis limits influenza virus, HIV, and dengue virus infection. Immunity 44, 46–58 (2016). SAINT analysis of AP-MS data . Two biological repeats were performed for each 8. Pichlmair, A. et al. Viral immune modulators perturb the human molecular IAV protein complex under IAV and mock infection conditions, respectively. The network by common and unique strategies. Nature 487, 486–490 (2012). resulting data are presented in Supplementary Data 1 and 6 and were compared 9. Mertens, C., Kuhn, C. & Franke, W. W. 2a and 2b: constitutive with 77 vector controls (cells transfected with empty vector), our published data- proteins of dual location in the karyoplasm and the desmosomal plaque. J. Cell base of 101 protein complexes12–14 and 60 unpublished protein complexes that 135, were purified from stable HEK293 cell lines expressing FLAG tag-fused non-viral Biol. 1009–1025 (1996). proteins handled in identical fashion (The whole data set used for SAINT analysis 10. Gerulll, B. et al. Mutations in the desmosomal protein plakophilin-2 are is included in Supplementary Data 3). Proteins found in the control group common in arrhythmogenic right ventricular cardiomyopathy. Nat. Genet. 36, were considered as non-specific binding proteins. The SAINT algorithm 1162–1164 (2004). (http://sourceforge.net/projects/saint-apms) was used to evaluate the MS data11. 11. Choi, H. et al. SAINT: probabilistic scoring of affinity purification-mass The SAINT utilizes label-free quantitative information to compute confidence spectrometry data. Nat. Methods 8, 70–73 (2011). scores (probability) for putative interactions11,51,52. Such quantitative information 12. Li, S., Wang, L., Berman, M., Kong, Y. Y. & Dorf, M. E. Mapping a dynamic can include counts (for example, spectral counts or number of unique peptides). In innate immunity protein interaction network regulating type I interferon an optimal setting, SAINT utilizes negative control immunoprecipitation data production. Immunity 35, 426–440 (2011). (typically, purifications without expression of the bait protein or with expression of 13. Li, S. et al. TRIM65 regulates microRNA activity by ubiquitination of TNRC6. an unrelated protein) to identify non-specific interactions in a semi-supervised Proc. Natl Acad. Sci. USA 111, 6970–6975 (2014). manner. A separate unsupervised SAINT modelling is capable of scoring 14. Fu, B., Li, S., Wang, L., Berman, M. A. & Dorf, M. E. The ubiquitin conjugating interactions in the absence of implicit control data, but only when a sufficient enzyme UBE2L3 regulates TNFalpha-induced linear ubiquitination. Cell Res. number of experiments are used for the modelling52. The default SAINT options 24, 376–379 (2014). were low Mode ¼ 1, min Fold ¼ 0 and norm ¼ 0. The SAINT scores computed for 15. Zeng, H., Pappas, C., Katz, J. M. & Tumpey, T. M. The 2009 pandemic H1N1 each biological replicate were averaged (AvgP) and reported as the final SAINT and triple-reassortant swine H1N1 influenza viruses replicate efficiently but

NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms13876

elicit an attenuated inflammatory response in polarized human bronchial in desmosomes and beta -catenin signaling. J. Biol. Chem. 277, 10512–10522 epithelial cells. J. Virol. 85, 686–696 (2011). (2002). 16. Hale, B. G., Jackson, D., Chen, Y. H., Lamb, R. A. & Randall, R. E. Influenza A 45. Calkins, C. C. et al. The armadillo family protein p0071 is a VE-cadherin- and virus NS1 protein binds p85beta and activates phosphatidylinositol-3-kinase -binding protein. J. Biol. Chem. 278, 1774–1783 (2003). signaling. Proc. Natl Acad. Sci. USA 103, 14194–14199 (2006). 46. Versteeg, G. A. et al. The E3-ligase TRIM family of proteins regulates signaling 17. York, A., Hutchinson, E. C. & Fodor, E. Interactome analysis of the influenza A pathways triggered by innate immune pattern-recognition receptors. Immunity virus transcription/replication machinery identifies protein phosphatase 6 as a 38, 384–398 (2013). cellular factor required for efficient virus replication. J. Virol. 88, 13284–13299 47. Liu, C. et al. Control of beta-catenin phosphorylation/degradation by a dual- (2014). kinase mechanism. Cell 108, 837–847 (2002). 18. Engelhardt, O. G., Smith, M. & Fodor, E. Association of the influenza A virus 48. Heaton, N. S. et al. In vivo bioluminescent imaging of influenza A virus RNA-dependent RNA polymerase with cellular RNA polymerase II. J. Virol. 79, infection and characterization of novel cross-protective monoclonal antibodies. 5812–5818 (2005). J. Virol. 87, 8272–8281 (2013). 19. Deng, T. et al. Role of ran binding protein 5 in nuclear import and assembly of 49. Szretter, K. J., Balish, A. L. & Katz, J. M. Influenza: propagation, quantification, the influenza virus RNA polymerase complex. J. Virol. 80, 11911–11919 (2006). and storage. Curr. Protoc. Microbiol. Chapter 15, Unit 15G 11 (2006). 20. Paoletti, A. C. et al. Quantitative proteomic analysis of distinct mammalian 50. Matrosovich, M., Matrosovich, T., Garten, W. & Klenk, H. D. New Mediator complexes using normalized spectral abundance factors. Proc. Natl low-viscosity overlay medium for viral plaque assays. Virol. J. 3, 63 (2006). Acad. Sci. USA 103, 18928–18933 (2006). 51. Breitkreutz, A. et al. A global protein kinase and phosphatase interaction 21. Das, K. et al. Structural basis for suppression of a host antiviral response by network in yeast. Science 328, 1043–1046 (2010). influenza A virus. Proc. Natl Acad. Sci. USA 105, 13093–13098 (2008). 52. Choi, H. et al. Analyzing protein-protein interactions from affinity 22. Popow, J., Jurkin, J., Schleiffer, A. & Martinez, J. Analysis of orthologous groups purification-mass spectrometry data with SAINT. Curr. Protoc. Bioinformatics reveals archease and DDX1 as tRNA splicing factors. Nature 511, 104–U508 Chapter 8, Unit8 15 (2012). (2014). 53. Ong, S. E. & Mann, M. A practical recipe for stable isotope labeling by amino 23. Heaton, N. S. et al. In vivo bioluminescent imaging of influenza a virus acids in cell culture (SILAC). Nat. Protoc. 1, 2650–2660 (2006). infection and characterization of novel cross-protective monoclonal antibodies. 54. Bradel-Tretheway, B. G. et al. Comprehensive proteomic analysis of influenza J. Virol. 87, 8272–8281 (2013). virus polymerase complex reveals a novel association with mitochondrial 24. Chan, S. L. et al. CPSF30 and Wdr33 directly bind to AAUAAA in mammalian proteins and RNA polymerase accessory factors. J. Virol. 85, 8569–8581 (2011). mRNA 30 processing. Genes. Dev. 28, 2370–2380 (2014). 55. Shapira, S. D. et al. A physical and regulatory map of host-influenza 25. Nemeroff, M. E., Barabino, S. M., Li, Y., Keller, W. & Krug, R. M. Influenza interactions reveals pathways in H1N1 infection. Cell 139, 1255–1267 (2009). virus NS1 protein interacts with the cellular 30 kDa subunit of CPSF and 56. Huang, D. W., Sherman, B. T. & Lempicki, R. A. Systematic and integrative inhibits 3’end formation of cellular pre-mRNAs. Mol. Cell 1, 991–1000 (1998). analysis of large gene lists using DAVID bioinformatics resources. Nat. Protoc. 26. Hale, B. G. et al. Inefficient control of host by the 2009 4, 44–57 (2009). pandemic H1N1 influenza A virus NS1 protein. J. Virol. 84, 6909–6922 (2010). 57. Orchard, S. et al. The MIntAct project--IntAct as a common curation platform 27. Bouwmeester, T. et al. A physical and functional map of the human for 11 molecular interaction databases. Nucleic Acids Res. 42, D358–D363 (2014). TNF-alpha/NF-kappa B signal transduction pathway. Nat. Cell Biol. 6, 97–105 (2004). Acknowledgements 28. Nourbakhsh, M. & Hauser, H. Constitutive silencing of IFN-beta promoter is mediated by NRF (NF-kappa B-repressing factor), a nuclear inhibitor of We thank Michael Berman for expert technical assistance, Drs Peter Palese (Mount Sinai NF-kappa B. Embo J. 18, 6415–6425 (1999). School of Medicine, NY, NY), Adolfo Garcia-Sastre (Mount Sinai School of Medicine, 29. Sun, H. H. et al. TIPE2, a negative regulator of innate and adaptive immunity NY, NY), Charles Russell (St Jude’s Hospital, Memphis, TN), Sean Whelan (Harvard that maintains immune homeostasis. Cell 133, 415–426 (2008). Medical School, Boston, MA), Robert Webster (St Jude Children’s Hospital, Memphis, 30. Tomita, T. et al. ZFC3H1, a zinc finger protein, modulates IL-8 transcription by TN), Ervin Fodor (University of Oxford, Oxford, UK), Andrew Rice (Baylor College binding with celastramycin A, a potential immune suppressor. PLoS ONE 9, of Medicine, Houston, TX), Koyu Hara (Kurume University School of Medicine, Japan) e108957 (2014). and Andrew Kowalczyk (Emory University, Atlanta, GA) for reagents. This work was 31. Momose, F. et al. Identification of Hsp90 as a stimulatory host factor involved supported by NIH grant R01AI089829 (M.D.), R01AI121288 (M.D. and S.L.), in influenza virus RNA synthesis. J. Biol. Chem. 277, 45306–45314 (2002). R01HL116876 (L.L.), R21AI121591 (L.L.), RAC Fund (S.L.) and NIH pilot grant 32. Hatzfeld, M. The armadillo family of structural proteins. Int. Rev. Cytol. 186, P20GM103648 (S.L.). 179–224 (1999). 33. Bass-Zubek, A. E., Godsel, L. M., Delmar, M. & Green, K. J. Plakophilins: Author contribution multifunctional scaffolds for adhesion and signaling. Curr. Opin. Cell Biol. 21, Conceived and designed the experiments: L.W., M.E.D., S.L. Performed the experiments: 708–716 (2009). L.W., B.F., W.L., G.P. Analysed the data: L.W., B.F., S.L. Wrote the paper: L.W., L.L., 34. Choi, H. J. & Weis, W. I. Structure of the armadillo repeat domain of M.E.D., S.L. plakophilin 1. J. Mol. Biol. 346, 367–376 (2005). 35. Mertens, C. et al. Nuclear particles containing RNA polymerase III complexes associated with the junctional plaque protein plakophilin 2. Proc. Natl Acad. Additional information Sci. USA 98, 7795–7800 (2001). Supplementary Information accompanies this paper at http://www.nature.com/ 36. Pflug, A., Guilligay, D., Reich, S. & Cusack, S. Structure of influenza A naturecommunications polymerase bound to the viral RNA promoter. Nature 516, 355 (2014). 37. Reich, S. et al. Structural insight into cap-snatching and RNA synthesis by Competing financial interests: The authors declare no competing financial interests. influenza polymerase. Nature 516, 361–366 (2014). 38. Obayashi, E. et al. The structural basis for an essential subunit interaction in Reprints and permission information is available online at http://npg.nature.com/ influenza virus RNA polymerase. Nature 454, 1127–U1157 (2008). reprintsandpermissions/ 39. Li, C. et al. A peptide derived from the C-terminus of PB1 inhibits influenza How to cite this article: Wang, L. et al. Comparative Influenza Protein Interactomes virus replication by interfering with viral polymerase assembly. FEBS J. 280, Identify the Role of Plakophilin 2 in Virus Restriction. Nat. Commun. 8, 13876 1139–1149 (2013). doi: 10.1038/ncomms13876 (2017). 40. Reuther, P., Manz, B., Brunotte, L., Schwemmle, M. & Wunderlich, K. Targeting of the influenza A virus polymerase PB1-PB2 interface indicates Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in strain-specific assembly differences. J. Virol. 85, 13298–13309 (2011). published maps and institutional affiliations. 41. Fodor, E. et al. Rescue of influenza A virus from recombinant DNA. J. Virol. 73, 9679–9682 (1999). 42. Hoffmann, E., Neumann, G., Kawaoka, Y., Hobom, G. & Webster, R. G. This work is licensed under a Creative Commons Attribution 4.0 A DNA transfection system for generation of influenza A virus from eight International License. The images or other third party material in this plasmids. Proc. Natl Acad. Sci. USA 97, 6108–6113 (2000). article are included in the article’s Creative Commons license, unless indicated otherwise 43. Nakazono, Y., Hara, K., Kashiwagi, T., Hamada, N. & Watanabe, H. The RNA in the credit line; if the material is not included under the Creative Commons license, polymerase PB2 subunit of influenza A/HongKong/156/1997 (H5N1) restrict users will need to obtain permission from the license holder to reproduce the material. the replication of reassortant ribonucleoprotein complexes. Plos One 7, e32634 To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/ (2012). 44. Chen, X., Bonne, S., Hatzfeld, M., van Roy, F. & Green, K. J. Protein binding and functional characterization of plakophilin 2. Evidence for its diverse roles r The Author(s) 2017

12 NATURE COMMUNICATIONS | 8:13876 | DOI: 10.1038/ncomms13876 | www.nature.com/naturecommunications